Enhancing Blind Image Deblurring Robustness Against Impulse Noise via Adaptive Noise Detection and Graph Regularization

Blind image deblurring is a challenging task aimed at recovering latent images from blurred observations. The presence of impulse noise significantly complicates this process by degrading image details and edge information, adversely affecting kernel estimation and image restoration. To address this issue, we propose a robust blind image deblurring algorithm that integrates adaptive impulse noise detection. Our approach employs an adaptive noise detection method to construct a noise mask matrix, which is then embedded into a graph-regularized blur kernel restoration model. This integration mitigates the detrimental effects of impulse noise on kernel estimation. Furthermore, we introduce a noise weight matrix correction term into the latent clean image restoration process to enhance the accuracy of noise handling. Experimental results demonstrate that our algorithm outperforms existing methods, achieving higher PSNR and SSIM values for restored images under various impulse noise densities. Our work not only advances the state-of-the-art in blind image deblurring but also provides a practical solution for handling impulse noise in real-world applications.The code files can be downloaded from "https://github.com/Edith0000/Enhancing-Blind-Image-Deblurring-Robustness-Against-Impulse-Noise".